Basic knowledge of database (interview)

One 、 Basic knowledge of

1、Char and Varchar difference ？

（1）Char It's fixed length , and Varchar Yes, it can grow .

Char Space will be allocated according to the declared string length , The right side of the string will be filled with spaces .

（2） In terms of storage mode ,Char Occupation of English characters 1 byte , Use... For a Chinese character 2 byte . and Varchar  Use... For each character 2 byte .

2、 What are the three paradigms of database ？

（1） First normal form ： Any table should have Primary key , And the atomicity of each field can no longer be divided .

（2） Second normal form ： Based on the First normal form Based on , All non primary key fields depend entirely on the primary key , Can't generate partial dependence .

（3） Third normal form ： Based on the second paradigm , All non primary key fields directly depend on the primary key , Cannot generate delivery dependency .

Many to many ？ Use three tables , Relation table two Foreign keys . 

One to many ？ Two tables , Add more watches Foreign keys .

Two 、 Indexes

1、 Classification of indexes

Single index ： Add an index to a single field

Composite index ： Add an index to the union of multiple fields

primary key ： An index is automatically added to the primary key

unique index ： Yes unique Indexes are automatically added to the constrained fields

2、 Advantages and disadvantages of index

An index is equivalent to a list of books , Through the directory, you can quickly find the corresponding resources .

In terms of databases , There are two retrieval methods when querying a table ： Full table scan 、 Index search （ It's very efficient ）.

Although index can improve retrieval efficiency , But you can't add indexes at will , Because the index is also an object in the database , It also requires continuous database maintenance . such as , The data in the table is often modified , This is not suitable for adding indexes , Because once the data is modified , The index needs to be reordered , For maintenance .

3、 The design principle of index （ When to consider adding an index to a field ）

The amount of data is huge （ According to the needs of customers , According to the online environment ）

This field is rarely used DML operation （ Because the field cannot be modified , Indexes also need to be maintained ）

This field often appears in where clause （ Which field do you often query based on ）

• Choose a unique index ;
• Index fields that are often used as query criteria ;
• Order for frequent need 、 Index fields for grouping and combining operations ;
• Limit the number of indexes ;
• Watch 不 Suggest index （ For example, the magnitude is within one million ）;
• all 量 Using data 量 Fewer indexes ;
• Delete indexes that are no longer or rarely used .

4、 The data structure of the index

The data structure of index is related to the implementation of specific storage engine ,MySQL What is commonly used in  Hash  and  B+ Trees Indexes .

• Hash Indexes The bottom is  Hash  surface , When querying, call Hash Function gets the corresponding key value （ Corresponding address ）, Then go back to the table and query to obtain the actual data .

• B+ Tree index The underlying implementation principle is multi-channel balanced lookup tree , Every query starts from the root node , Query the leaf node to get the key value , Finally, query to determine whether it is necessary to query back to the table .

（1）Hash and B+ The difference between tree indexes

Hash
1）Hash Faster equivalent query , But you can't do a range query . Because after Hash After the function is indexed , The order of indexes cannot be consistent with the original order , Therefore, range query cannot be supported . Empathy , Nor does it support sorting by index .

2）Hash Fuzzy query and leftmost prefix matching of multi column index are not supported , because Hash The value of the function is unpredictable , Such as AA and AB There is no correlation between the calculated values of .

3）Hash You can't avoid querying data back to the table at any time .

4） Although the query efficiency is high in equivalence , But the performance is unstable , Because when there are a lot of duplicates in a key value , produce Hash Collision , At this time, the query efficiency may be reduced .

B+ Tree

1）B+ The essence of a tree is a search tree , Natural support range query and sorting .

2） When certain conditions are met （ Cluster index 、 Overwrite index, etc ） When you can complete the query only through the index , There is no need to return the form .

3） The query efficiency is relatively stable , Because every query is from the root node to the leaf node , And is the height of the tree .

（2） Why use B+ Trees Instead of Binary tree and B Trees Do the index

Binary tree

a.  If there is a lot of index data , The level of the tree will be very high （ Only the left and right child nodes ）, When the amount of data is large, the query will still be slow , The search efficiency is O(logn).

b. A binary tree stores only one record per node , When a query is found in the tree, it takes disk IO More times . When the file system needs to read data from disk , Generally read on a page basis , Suppose there is too little data in a page , Then the operating system needs to read more pages , Involving random disks I/O More visits . Reading data from disk into memory involves random I/O The interview of , It is one of the most expensive operations in the database .

B Trees

a. It's not a binary search anymore , It is N Cross search , The height of the tree will decrease , Quick query

b. Leaf node , Nonleaf node , Can store data , And can store multiple data

c. Traverse through the middle order , You can access all nodes in the tree

B+ Trees

B Both non leaf nodes and leaf nodes store data , So when querying data , The best time complexity is O(1), The worst is O(log n). and B+ Trees store data only at leaf nodes , Non leaf nodes store keywords , And the keywords of different non leaf nodes may be repeated , So when querying data , The time complexity is fixed to O(log n).

B+ The leaf nodes of trees are connected with each other by linked list , Therefore, only scanning the linked list of leaf nodes can complete a traversal operation ,B Trees can only be traversed through the middle order .

Why? B+ Tree ratio B Trees are more suitable for database indexes ？

B+ There are fewer trees IO frequency .
Because the index file is very large, the index file is stored on disk ,B+ The non leaf nodes of the tree only store keywords, not data , As a result, a single page can store more keywords , That is to say, the more keywords need to be searched when reading into memory at one time , Random disk I/O The number of reads is relatively reduced . and B Both non leaf nodes and leaf nodes store data .

B+ Tree query efficiency is more stable
Because the data only exists on the leaf node , So the search efficiency is fixed to O(log n), therefore B+ The query efficiency of tree is compared with B Trees are more stable .

B+ The tree is more suitable for range lookup
B+ The leaf nodes of a tree are connected in order by a linked list , So to scan all the data, you only need to scan the leaf node once , Easy to scan database and range query ;B Trees also store data because they are not leaf nodes , Therefore, we can only scan in order by traversing the middle order . in other words , For range queries and ordered traversal ,B+ More efficient trees .

3、 ... and 、 Storage

1、 Storage engine （MyISAM and InnoDB）

（1）InnoDB Support transactions , and MyISAM I won't support it .

（2）InnoDB Support foreign keys , and MyISAM I won't support it . So you're going to have a with a foreign key InnoDB surface To MyISAM The table will fail .

（3）InnoDB and MyISAM support B+ Tree Index of data structure . but InnoDB yes Clustered index , and MyISAM Yes no clustered index .

（4）InnoDB Do not save the number of data rows in the table , perform select count(*) from table You need a full scan . and MyISAM Use a variable to record the number of rows in the whole table , Pretty fast （ Be careful not to have WHERE Clause ）.

（5）InnoDB Support surface 、 That's ok （ Default ） Level lock , and MyISAM Support table level lock .

（6）InnoDB There has to be unique index （ Such as primary key ）, If not specified , Will automatically find or produce a hidden column Row_id To act as the default primary key , and MyISAM  There can be no primary key .

By default InnoDB,MyISAM It is applicable to the program dominated by insertion , Like the blog system 、 News portal .

2、 Storage structure

InnoDB Page of 、 District 、 paragraph ？

page （Page）
First ,InnoDB Divide the physical disk into pages （page）, The default size of each page is 16 KB, Page is the smallest storage unit .

District （Extent）
If there is only one level of page , The number of pages is very large , The allocation and recycling of storage space will be troublesome , Because maintaining the state of so many pages is very troublesome . therefore ,InnoDB It also introduces the area （Extent) The concept of . A zone defaults to 64 Consisting of consecutive pages , That is to say 1MB. adopt Extent It is easier to allocate and recycle storage space .

paragraph （Segment）
B+ The leaf node of the tree stores our specific data , Non leaf nodes are index pages . therefore B+ The tree divides the data into two parts , Leaf node part and non leaf node part , That's the paragraph we're going to introduce Segment, in other words InnoBD Two are created for each index in Segment To store the corresponding two parts of data .

Four 、 Business

1、 What is a database transaction ？

A transaction is a complete business logic unit , Can not be further divided .

Two of the above DML Statements must both succeed , Or fail at the same time , A success message is not allowed , One failure .

To ensure the above two DML Statements succeed or fail at the same time , Then you need to use the database “ Transaction mechanism ”.

2、 What are the four characteristics of a transaction ？（ACID）

A： Atomicity ： Transactions are the smallest unit of work , Can not be further divided .

C： Uniformity ： Transactions must guarantee multiple DML Statements succeed or fail at the same time .

I： Isolation, ： Business A And business B There is isolation between , The database between concurrent transactions is independent .

D： persistence ： The final data must be persisted to the hard disk file , The transaction is a successful end .

3、 Concurrency of transactions ？

Dirty reading ： One transaction reads uncommitted data from another transaction .  Business A Read transactions B Updated data , then B Rollback operation , that A The data read is dirty .

Fantasy reading ： The amount of data read twice in a transaction is inconsistent . System administrator A Change the scores of all students in the database from specific scores to ABCDE Grade , But the system administrator B At this time, a specific score record was inserted , When the system administrator A After the change, I found that there is another record that hasn't been changed , It's like an illusion , This is called Unreal reading .

It can't be read repeatedly ： The contents of the data read twice in a transaction are inconsistent . Business A Read the same data multiple times , Business B In the transaction A During multiple reads , The data has been updated and submitted , Cause transaction A When reading the same data multiple times , result atypism .

4、 The isolation level of the transaction

First level ： Read uncommitted （Read uncommitted）

The other party's transaction has not been committed , Our current transaction can read uncommitted data from the other party .

Read uncommitted, dirty read exists （dirty read） The phenomenon ： Indicates that dirty data has been read .

The second level ： Read submitted （Read committed）

We can read the data after the other party's transaction is committed .

The problem with reading submitted is ： It can't be read repeatedly .

The third level ： Repeatable （Repeatable read）

This isolation level solves the problem of non repeatable reads .

The problem with this level of isolation is ： The data read is an illusion 、

Level 4 ： Serializable （Serializable）

All problems solved . Low efficiency , Transaction queuing is required .

oracle The default isolation level of the database is read committed ;mysql database The default isolation level is Repeatable .

5、 ... and 、 lock

1、 The function and classification of database lock

When there are concurrent transactions in the database , There may be data inconsistencies , At this time, we need some mechanisms to ensure the order of access , Lock mechanism is such a mechanism . That is, the function of lock is to solve the problem of concurrency .

From the granularity of lock , Locks can be divided into table locks 、 Row lock and page lock

Row-level locks ： It is the lock with the smallest granularity , Indicates that only the row of the current operation is locked . Row level lock can greatly reduce the conflict of database operation . The lock granularity is the smallest , But locking costs the most .
The overhead of row level locking is large , Lock the slow , And there will be a deadlock . But the locking granularity is the smallest , The lowest probability of lock collisions , The highest degree of concurrency .

Table lock ： It is a lock with the largest granularity , Indicates to lock the whole table of the current operation , It's easy to implement , Less resource consumption , By most MySQL Engine support .

Page level lock ： It is a kind of lock with granularity between row level lock and table level lock . The watch lock is fast , But there are many conflicts , Less line level conflict , But the speed is slow. . So I took the middle page level , Lock an adjacent set of records at a time .

From the nature of use , Can be divided into shared locks 、 Exclusive lock and update lock

Shared lock ：S lock , Also called read lock , For all read-only data operations .
S Locks are not exclusive , Allow multiple concurrent transactions to lock the same resource , But add S It is not allowed to add... While locking X lock , That is, resources cannot be modified .S The lock is usually released immediately after reading , No need to wait for the transaction to end .

Exclusive lock ：X lock , Also known as write lock , Means to write data .
X Locks allow only one transaction to lock the same resource , And not released until the end of the transaction , Any other business has to wait until X The lock is released to access the page .

Update lock ：U lock , Used to book the resources to be imposed X lock , Allow other transactions to read , But no more U Lock or X lock .
When the page being read is about to be updated , Upgrade to X lock ,U The lock cannot be released until the end of the transaction . so U Locks are used to avoid deadlocks caused by shared locks .

Subjectively divide , It can also be divided into optimistic lock and pessimistic lock

Optimism lock ： Subjectively determine that resources will not be modified , So read data without locking , Only when updating, use the version number mechanism to confirm whether the resource has been modified .
Optimistic lock is suitable for multi read applications , The system throughput can be improved .

Pessimistic locking ： With strong exclusivity and exclusivity , Every time the data is read, it is considered that it will be modified by other transactions , So every operation needs to be locked .

2、 Relationship between isolation level and lock ？
1） stay Read Uncommitted Below grade , Reading data does not require a shared lock , This will not conflict with the exclusive lock on the modified data ;

2） stay Read Committed Below grade , Read operations require shared locks , But release the shared lock after the statement is executed ;

3） stay Repeatable Read Below grade , Read operations require shared locks , But the shared lock is not released before the transaction is committed , In other words, the shared lock must be released after the transaction is completed ;

4） stay SERIALIZABLE Below grade , The most restrictive , Because this level locks the entire range of keys , And hold the lock all the time , Until the transaction is complete .
3、 Snapshot read and current read

Read the snapshot Is to read snapshot data , Simple without lock Select All belong to snapshot reading .

The current reading Is to read the latest data , Not historical data . The lock SELECT, Or add, delete and modify the data, which will be read currently .

4、 What is? MVCC And realize ？

MVCC Multi version concurrency control , You can read and write without blocking each other , It is mainly used to improve the concurrency efficiency when solving the problems of non repeatable reading and unreal reading .

Its principle is to realize the concurrency control of database through multiple version management of data rows , Simply put, it's saving historical versions of the data . You can determine whether the data is displayed by comparing the version numbers . When reading data, there is no need to lock to ensure the isolation effect of transactions .

6、 ... and 、SQL sentence

1、SQL Left connection of 、 The right connection 、 Internal connection

（1）​Left join： That's left connection , It's based on the left table , according to ON The conditions of the two tables given later connect the two tables . The result will list all the query information in the left table , The right table only lists ON The latter condition is the same as the part satisfied by the left table . The full name of the left connection is the left outer connection , It is a kind of external connection .

（2）​Right join： Right connection , Is based on the right table , according to ON The conditions of the two tables given later connect the two tables . The result will list all the query information in the right table , The left table only lists ON The latter condition is the same as the part satisfied by the right table . The full name of right connection is right outer connection , It is a kind of external connection .​

（3）Inner join： Internal connection , At the same time, take the two tables as reference objects , according to ON The conditions of the two tables given later connect the two tables . The result is that both tables meet ON The following conditions will be listed .

The internal connection can only query the data that can be matched by two tables , Unmatched data cannot be found .

The external connection can unconditionally query all the data in the main table .

2、Where and Having The difference between ？

where Clause is used before grouping query results , Remove the unqualified lines , That is to say Filter data before grouping ,where A condition cannot contain a grouper function , Use where Conditions filter out specific lines .

having Clause is used to filter groups that meet the conditions , That is to say Filter data after grouping , Conditions often include groupings , Use having Conditions filter out specific groups , You can also use multiple grouping criteria for grouping .